Despite significant advances, artificial intelligence (AI) faces several fundamental limitations when analyzing underwater images for marine research, many of which stem directly from the unique and challenging nature of the marine environment. Unlike land-based imagery, underwater footage is often characterized by poor visibility, distorted colors, and a dynamic, unpredictable setting that can mislead even the most sophisticated AI algorithms.
Environmental distortions
Poor image quality: The ocean's properties—specifically the absorption and scattering of light—are the single greatest challenge for underwater AI. This leads to images with low contrast, reduced resolution, and a color cast (often blue or green) that varies with depth. These factors can create blurry images and obscure details, making accurate object detection and classification extremely difficult.
Irregular lighting and shadows: The availability and quality of light underwater are highly variable. Images can have uneven illumination, harsh shadows, and intense backscatter from artificial light sources. This inconsistency makes it hard for AI models to generalize, as they might struggle to perform accurately in a different location or time of day.
Data-related challenges
Insufficient labeled data: For AI models to learn to identify marine species, they need large, diverse datasets of expertly annotated images. However, obtaining and labeling high-quality underwater imagery is expensive, time-consuming, and requires specialized taxonomic knowledge. Many existing datasets are relatively small, lack diversity, or are biased towards common species, hindering a model's ability to accurately identify rare or new organisms.
High class imbalance: In any given underwater habitat, a few species are common, while many more are rare. AI training data reflects this reality, with some classes heavily overrepresented while others are severely underrepresented. This imbalance can lead to biased models that perform poorly when asked to identify a rare species, impacting conservation efforts that depend on monitoring such populations.
Domain shift: AI models trained on data from one region of the ocean often struggle to perform in another due to significant differences in environmental conditions, lighting, and species composition. A model trained on a clear, shallow coral reef may perform poorly in a turbid, deep-sea environment, creating a major generalization challenge for marine biologists.
Object-related complications
Camouflage and mimicry: Many marine species have evolved sophisticated camouflage to blend in with their surroundings. This makes them inherently difficult for AI to detect and identify, as the models must be able to discern subtle features that distinguish the organism from the background.
Occlusions and overlapping objects: Underwater scenes are often cluttered, with species partially or fully hidden by vegetation, coral, or other marine life. For example, schools of fish can overlap, making it challenging for AI to count and identify individuals accurately.
Scale variation: Marine organisms come in a massive range of sizes, from microscopic plankton to large whales. A single AI model may struggle to accurately detect and classify objects across such a vast scale, with smaller objects often being overlooked due to insufficient image resolution.
Algorithmic and practical limitations
Interpretability and trust: The "black box" nature of many deep learning models makes it difficult for marine biologists to understand and trust how the AI arrives at a specific conclusion. Without clear explanations, researchers may be hesitant to rely on AI-generated data for high-stakes decisions, such as directing conservation efforts.
Computational cost: Training complex AI models requires significant energy and computational power, which can be a limiting factor, especially when using underwater robots with restricted computing resources. Running the models in real-time, as required for autonomous research, remains computationally challenging.